In the moving picture coding process, the amount of data is compressed using redundancy of a moving picture in a spatial direction and a temporal direction in general. A transformation into a frequency domain is generally used as a method for using the redundancy in the spatial direction. In addition, as a way to use the redundancy in the temporal direction, an inter-picture prediction (hereafter referred to as an inter prediction) coding is used (for example, see the Non-Patent Literature 1).